The Startup Defending Us From AI-Designed Plagues

TL;DR

• 1

  An 80-fold lab improvement was the wake-up call — Rajaniemi says reasoning models helped invent a new molecular cloning protocol that beat Gibson assembly by roughly 79-80x, convincing him that AI was crossing from biotech assistant to genuine research engine.

• 2

  The same AI that helped on cancer vaccines also revealed a frightening dual use — After OpenAI’s o3 zero-shotted a three-mRNA cancer-immunotherapy cocktail his team had spent 18 months and 20 iterations developing, Red Queen’s founders explored the “dark twin” of that capability and decided biodefense had become urgent.

• 3

  Red Queen Bio’s thesis is ‘new pathogen in, countermeasures out’ — The company wants near-zero design time for therapies and prophylactics, using AI for antibody design, biological reasoning, and fast manufacturing so outbreaks can be ring-contained the way smallpox and Ebola responses were.

• 4

  The near-term threat isn’t necessarily a super-pandemic tomorrow — Rajaniemi worries first about escalating localized incidents from “doomsday cults” and lone actors getting AI uplift, with attempts becoming more frequent and more successful before truly catastrophic engineered outbreaks arrive.

• 5

  He thinks AI alignment needs politics, not just engineering — Rajaniemi argues society needs broader deliberation, including citizen assemblies, because decisions about AI constitutions, regulation, and human-AI relationships can’t be left to a small circle of San Francisco labs.

• 6

  His sci-fi frame for AI is closer to Frankenstein than Terminator — The lesson of Frankenstein, he says, is not “don’t create life,” but “don’t abandon what you create,” and that treating AI only as a caged demon may help produce exactly that outcome.